Likelihoods for a general class of ARGs under the SMC

Area of Science:

• Population Genetics
• Computational Biology
• Bioinformatics

Background:

• Ancestral recombination graphs (ARGs) are crucial for understanding genetic variation.
• Recent advances enable ARG inference for large sample sizes, but heuristic methods lack topological accuracy for likelihood computation.
• Current methods struggle with precise recombination event details needed for models like the Sequentially Markov Coalescent (SMC).

Purpose of the Study:

• To present a backwards-time formulation of the SMC model.
• To derive a straightforward definition of ARG likelihood under this model.
• To develop ARG inference methods robust to recombination event precision and polytomies.

Main Methods:

• Developed a backwards-time formulation of the SMC model.
• Derived a likelihood definition for ARGs under this formulation.
• Investigated robustness to polytomies and imprecise recombination event data.

Main Results:

• The new formulation allows ARG likelihood computation without precise recombination event details.
• The method is robust to the presence of polytomies in ARGs.
• This work facilitates improved ARG inference, especially for large datasets.

Conclusions:

• The backwards-time SMC formulation simplifies ARG likelihood calculation.
• This approach enhances the feasibility and accuracy of ARG inference.
• Opens new possibilities for scalable and robust ARG inference in population genetics.